🧠 Advanced Computer Vision - Class Activation Mapping (CAM) with TensorFlow

This repository contains two mini-projects demonstrating Class Activation Mapping (CAM) using pre-trained CNNs on:

1. 🐱🐶 Cats vs. Dogs Dataset
2. 👕👖 Fashion MNIST Dataset

Both notebooks visualize model attention using Grad-CAM to better understand CNN decision-making in image classification tasks.

---

📂 Contents

• CatsDogs_CAM.ipynb: Visual explanation of binary classification using MobileNetV2 and Grad-CAM on Cats vs. Dogs.
• FashionMNIST_CAM.ipynb: Grad-CAM on FashionMNIST with a custom CNN to inspect feature relevance.

---

🔍 Project Objective

To demonstrate interpretability in CNNs using Class Activation Maps, helping users:

• Visualize where the model is "looking" when classifying images
• Gain intuition about model trust and errors
• Learn to implement Grad-CAM from scratch using TensorFlow

---

🛠️ Technologies Used

• Python 3.10+
• TensorFlow 2.x / Keras
• NumPy, Matplotlib, OpenCV
• Grad-CAM (custom logic)

---

📚 Datasets

• 🐾 Cats vs. Dogs
  From TensorFlow Datasets (TFDS)
  Binary classification between cat and dog images.

• 🛍 Fashion MNIST
  28x28 grayscale images of clothing items
  Multi-class classification with 10 labels

---

▶️ Getting Started

1. Clone the Repo

git clone https://github.com/yourusername/vision-cam-tensorflow.git
cd vision-cam-tensorflow